Lake outbursts

The catastrophic release of impounded water in smallish glacial lakes creates medium-scale geomorphic features. The Watrous spillway, Saskatchewan, Canada, rapidly incised during a short-lived outburst from glacial Lake Elstow (Kehew and Teller 1994). In its outlet area, the bed of Lake Elstow was composed of stagnant ice. The 40-km-long spillway cuts across a divide, and ends in the glacial Last Mountain Lake basin, where a coarse-grained fan was deposited. Large clasts are concentrated on the fan surface, and probably represent deposition at peak discharge. Such outburst features are uncontroversial. More contentious is the suggestion that very large bodies of water impounded by ice or sediment cause catastrophic floods if suddenly freed. Such speculative events were ridiculed when first proposed, but, in the face of overwhelming evidence, they are widely accepted today (see

Huggett 1989b, 149-59). Three cases will illustrate the catastrophic nature of such floods: Pleistocene floods in the south-eastern Washington State, Pleistocene floods in Russia, and jokulhlaups.

Ice age floods in the Pacific Northwest

The Spokane Flood is the prime example of several well-documented cases of ice-dam breakage (p. 5). It took place between 13,000-18,000 years ago in southeastern Washington State and involved two outbursts from Glacial Lake Missoula following the failure of impounding dams of ice that created the Channeled Scablands (Baker 1978a, 1978b) (Figure 5.2). Further studies have shown that, during Quaternary times, at least five major cataclysmic floods occurred in the general vicinity of the Channeled Scablands, of which the Spokane Flood was the last. Secondary advances of the Cordilleran ice sheet during the Wisconsin and during previous major glacial cycles would have provided opportunities for ice-dammed lakes to form. Earlier outburst floods occurred in the Pacific Northwest, though erosion and deposition by younger floods and the deposition of extensive Holocene loess tend to obscure evidence of their existence (Bjornstad et al. 2001). An evaluation of surface exposures and borehole studies suggest that they started around 1.5 to 2.5 million years ago. At least two episodes of pre-Wisconsin catastrophic glacial-outburst flooding occurred - a Middle Pleistocene flood (older than 130,000 years) and an Early Pleistocene flood (older than 780,000 years). Surface exposures, using radiometric age dates, and palaeomagnetic and pedogenic evidence, were used to identify these floods. Exposures of pre-Wisconsin flood deposits on the Channeled Scabland are scarce owing to erosion of the flood-scoured coulees by later floods. However, in depositional basins beyond the Channeled Scabland, flood deposits up to 100 m thick accumulated behind hydraulically dammed constrictions along the flood path. Unlike surface exposures, flood bars within these depositional basins furnish a longer-term and more complete record of earlier Pleistocene flood episodes. These bars grew piecemeal and they represent an amalgamation of cataclysmic flood deposits laid down intermittently through the Pleistocene. In one giant flood bar, up to 100 m thick, deposits interpreted as Matuyama age indicate that the bar had grown to half its present height by 780,000 years ago. Additionally, Matuyama-age, reversed-polarity flood deposits may overlie up to another 15 m of normally magnetized deposits at the base of the flood sequence. This normal-polarity interval appears to be associated with Early Pleistocene cataclysmic floods, perhaps of Olduvai age (older than 1.77 million years). It is possible that many of the features associated with cataclysmic floods - coulees, giant bars, and streamlined loess hills, for instance - formed during the Early Pleistocene, and then suffered only slight alteration by up to hundreds of subsequent flood episodes.

Other catastrophic lake bursts occurred in North America. They include the Lake Bonneville Flood (Malde 1968; Jarrett and Malde 1987), which took place about 15,000 years ago, and the catastrophic drainage of glacial Lake Agassiz through a northwestern outlet following the incision of a drainage divide about 9,900 years ago (Smith and Fisher 1993). Pleistocene Lake Bonneville overtopped its Rim at Red Rock Pass in south-eastern Idaho and rapidly lowered, decanting about 4,700 km3 of water down the Snake River (Malde 1968). This debacle rushed down the Snake River Plain of southern Idaho to Hell's Canyon, causing extensive erosion and deposition. Today, the valley displays impressive abandoned channels, areas of scabland, and gravel bars composed of sand and angular and rounded boulders up to 3 m in diameter. The peak discharge, calculated using a step-back-

Figure 5.2 Glacial Lake Missoula and the Channeled Scabland. Two outbursts from glacial Lake Missoula, which took place between 18,000 and 13,000 years ago, produced massive floods in southeastern Washington State. Evidence of these debacles includes abandoned waterways, cataract cliffs and plunge basins, potholes and deep rock basins, giant bars and giant ripples. Source: Adapted from Baker (1978a).

Figure 5.2 Glacial Lake Missoula and the Channeled Scabland. Two outbursts from glacial Lake Missoula, which took place between 18,000 and 13,000 years ago, produced massive floods in southeastern Washington State. Evidence of these debacles includes abandoned waterways, cataract cliffs and plunge basins, potholes and deep rock basins, giant bars and giant ripples. Source: Adapted from Baker (1978a).

water computational technique for the constricted reach of the Snake River Canyon at the mouth of Sinker Creek, was 793,000-1,020,000 m3/s (Jarrett and Malde 1987). At this rate of discharge, the shear stress for the flood would have been 2,500 N/m2 and the unit stream power would have been 75,000 N/m-s. This compares with shear stress and unit stream power for recent floods of the Mississippi and Amazon rivers of 6-10 N/m2 and 12 N/m-s.

Outburst floods in Russia

The Altai Mountains in southern Russia consist of huge intermontane basins and high mountain ranges, some over 4,000 m. During the Pleistocene, the basins contained lakes wherever glaciers grew large enough to act as dams. Research in this remote area has revealed a fascinating geomorphic history (Rudoy 1998). The glacier-dammed lakes regularly burst out to generate glacial superfloods that have left behind exotic relief forms and deposits - giant current ripple-marks, diluvial swells and terraces, spillways, outburst and oversplash gorges, dry waterfalls, and so on. These features are allied to the Channeled Scabland features of Washington State, USA, which were produced by catastrophic outbursts from glacial lake Missoula. The outburst superfloods discharged at a rate in excess of 106 m3/s, flowed at dozens of metres a second, and some stood more than one hundred metres deep. The superpowerful diluvial waters changed the land surface in minutes, hours, and days. Diluvial accumulation, diluvial erosion, and diluvial evorsion were widespread. Diluvial accumulation built up ramparts and terraces (some of which were made of deposits

240 m thick), diluvial berms (large-scale counterparts of boulder-block ramparts and spits - 'cobblestone pavements' - on big modern rivers), and giant ripple-marks with wavelengths up to 200 m and heights up to 15 m. Some giant ripple-marks in the foothills of the Altai, between Platovo and Podgornoye, which lie 300 km from the site of the flood outbursts, point to a mean flood velocity of 16 m/s, a flood depth of 60 m, and a discharge of no less than 600,000 m3/s. Diluvial supererosion led to the formation of deep outburst gorges, open-valley spillways, and diluvial valleys and oversplash gorges where water could not be contained within the valley and plunged over the local watershed. Diluvial evorsion, which occurred beneath mighty waterfalls, forced out hollows in bedrock that today are dry or occupied by lakes.


These are outbursts of meltwater stored beneath a glacier or ice sheet as a subglacial lake. These best-known jokulhlaups occurred in the last century, with major ones in 1918 (Katla) and 1996 (Skeidararsandur) (Gudmundsson et al. 1995). Evidence of jokulhlaups during the Pleistocene exists (Geirsdottir et al. 2000; Mokhtari Fard and Ringberg 2001). Skeidararsandur jokulhlaup resulted from the rapid melting of some 3.8 km3 of ice (Russell et al. 1999) after a volcanic eruption on 30 September 1996 underneath the Vatnajokull ice cap (Gudmundsson et al. 1997). The ensuing flood involved a discharge of about 20,000 m3/s, running at its peak at around 6 m/s and capable of transporting ice blocks at least 25 m large (van Loon 2004). It destroyed part of the main road along the southern coast of Iceland, including a bridge over the Skeidararsandur. Catastrophic though the Skeidararsandur jokulhlaup was, it was tame in comparison with the 1918 Katla jokulhlaup, which involved a flood of about 300,000 m3/s of water that carried 25,000 tons of ice and an equal amount of sediment every second (Tomasson 1996).

Much larger jokulhlaups seem possible. Lake Vostok, lying beneath the Antarctic ice sheet, holds 5,000km3 or water and covers an area of some 14,000 km2 (roughly 200 x 70 km) (Siegert 2000). Tom van Loon (2004) speculates on the consequences of 10 per cent of Lake Vostok suddenly releasing. Its drainage rate would match Lake Missoula's peak discharge, and last eight days. A flood of that magnitude and length would exert huge forces on the roofs and the walls of the subglacial tunnels. Moreover, the 'heat' transport would be large enough to destroy large parts of the roofs and walls by thermo-erosion, which would destabilize ice around the subglacial tunnels. Water of that volume would be unlikely to drain through a single outlet; more likely, it would force other outlets through stress-induced crevasses. With a crevasse network formed in the lower part of the ice mass, a growing portion of the ice would come to rest on a rapidly moving layer of lubricating water. Near ice front, principally where the ice is in direct contact with the ocean, the lubricating water layer would aid the formation of ice masses 'floating' over the flood in a downslope direction (a slope of less than 1° would suffice), so causing icebergs to calve into the ocean. The sliding ice masses might originate at vertical faults in the ice resulting from shock waves created by the sudden pressure exerted on subglacial ice walls by the surging floodwater. The larger the flood, the greater the destruction of the ice mass, owing to the positive relationship between the force of a subglacial jokulhlaup (and thus the discharge) and the distance from the ice front where cracks may form. In consequence, giant jokulhlaups might set off the release of giant ice masses. Heinrich events (the calving of ice masses from the Laurentide ice sheet large enough to leave traces of their southward transport across the Atlantic Ocean) might result.

Was this article helpful?

0 0

Post a comment